--- license: mit tags: - adafactor - co-attention - deit - instancenorm - linear - onecycle - retrieval - swish - tiny - trunc-normal --- # model_562734402_deit_tiny.py ## Model Overview A **tiny**-scale implementation of the **deit** architecture, built for **retrieval** tasks. ## Architecture - **Architecture**: deit - **Scale**: tiny - **Attention**: linear - **Fusion strategy**: co attention - **Task head**: retrieval - **Activation**: swish - **Normalization**: instancenorm - **Initialization**: trunc normal ## Training - **Optimizer**: adafactor - **LR scheduler**: onecycle ## Files - `model_562734402_deit_tiny.py` — main artifact of this repository ## License See the license field above.